Coordinated campaigns are used to manipulate social media platforms and influence their users, a critical challenge to the free exchange of information. Our paper introduces a general, unsupervised, network-based methodology to uncover groups of accounts that are likely coordinated. The proposed method constructs coordination networks based on arbitrary behavioral traces shared among accounts. We present five case studies of influence campaigns, four of which in the diverse contexts of U.S. elections, Hong Kong protests, the Syrian civil war, and cryptocurrency manipulation. In each of these cases, we detect networks of coordinated Twitter accounts by examining their identities, images, hashtag sequences, retweets, or temporal patterns. The proposed approach proves to be broadly applicable to uncover different kinds of coordination across information warfare scenarios.
By Diogo Pacheco, Pik-Mai Hui, Chris Torres, Bao Truong, Sandro Flammini & Fil Menczer
Read the full open-access article from the Proceedings ICWSM2021
Kazutoshi Sasahara, Wen Chen, Hao Peng, Giovanni Luca Ciampaglia, Alessandro Flammini & Filippo Menczer
Journal of Computational Social Science (2020)
While social media make it easy to connect with and access information from anyone, they also facilitate basic influence and unfriending mechanisms that may lead to segregated and polarized clusters known as “echo chambers.” Here we study the conditions in which such echo chambers emerge by introducing a simple model of information sharing in online social networks with the two ingredients of influence and unfriending. Users can change both their opinions and social connections based on the information to which they are exposed through sharing. The model dynamics show that even with minimal amounts of influence and unfriending, the social network rapidly devolves into segregated, homogeneous communities. These predictions are consistent with empirical data from Twitter. Although our findings suggest that echo chambers are somewhat inevitable given the mechanisms at play in online social media, they also provide insights into possible mitigation strategies.
Saad Alqithami, Rahmat Budiarto, Musaad Alzahrani and Henry Hexmoor
Entropy 2020, 22(5), 519
Due to the complexity of an open multi-agent system, agents’ interactions are instantiated spontaneously, resulting in beneficent collaborations with one another for mutual actions that are beyond one’s current capabilities. Repeated patterns of interactions shape a feature of their organizational structure when those agents self-organize themselves for a long-term objective. This paper, therefore, aims to provide an understanding of social capital in organizations that are open membership multi-agent systems with an emphasis in our formulation on the dynamic network of social interactions that, in part, elucidate evolving structures and impromptu topologies of networks. We model an open source project as an organizational network and provide definitions and formulations to correlate the proposed mechanism of social capital with the achievement of an organizational charter, for example, optimized productivity. To empirically evaluate our model, we conducted a case study of an open source software project to demonstrate how social capital can be created and measured within this type of organization. The results indicate that the values of social capital are positively proportional towards optimizing agents’ productivity into successful completion of the project.
Mohsen Mosleh, Gordon Pennycook, David G. Rand
There is an increasing imperative for psychologists and other behavioral scientists to understand how people behave on social media. However, it is often very difficult to execute experimental research on actual social media platforms, or to link survey responses to online behavior in order to perform correlational analyses. Thus, there is a natural desire to use self-reported behavioral intentions in standard survey studies to gain insight into online behavior. But are such hypothetical responses hopelessly disconnected from actual sharing decisions? Or are online survey samples via sources such as Amazon Mechanical Turk (MTurk) so different from the average social media user that the survey responses of one group give little insight into the on-platform behavior of the other? Here we investigate these issues by examining 67 pieces of political news content. We evaluate whether there is a meaningful relationship between (i) the level of sharing (tweets and retweets) of a given piece of content on Twitter, and (ii) the extent to which individuals (total N = 993) in online surveys on MTurk reported being willing to share that same piece of content. We found that the same news headlines that were more likely to be hypothetically shared on MTurk were also shared more frequently by Twitter users, r = .44. For example, across the observed range of MTurk sharing fractions, a 20 percentage point increase in the fraction of MTurk participants who reported being willing to share a news headline on social media was associated with 10x as many actual shares on Twitter. We also found that the correlation between sharing and various features of the headline was similar using both MTurk and Twitter data. These findings suggest that self-reported sharing intentions collected in online surveys are likely to provide some meaningful insight into what content would actually be shared on social media.
Demival Vasques Filho, Dion R. J. O’Neale
Dynamical processes, such as the diffusion of knowledge, opinions, pathogens, "fake news", innovation, and others, are highly dependent on the structure of the social network on which they occur. However, questions on why most social networks present some particular structural features, namely high levels of
transitivity and degree assortativity, when compared to other types of networks remain open. First, we argue that every one-mode network can be regarded as a projection of a bipartite network, and show that this is the case using two simple examples solved with the generating functions formalism. Second, using synthetic and empirical data, we reveal how the combination of the degree distribution of both sets of nodes of the bipartite network — together with the presence of cycles of length four and six — explains the observed levels of transitivity and degree assortativity in the one-mode projected network. Bipartite networks with top node degrees that display a more right-skewed distribution than the bottom nodes result in highly transitive and degree assortative projections, especially if a large number of small cycles are present in the bipartite structure.